12
Conformational Diversity of the Helix 12 of the Ligand Binding Domain of PPARγ and Functional Implications Mariana R. B. Batista and Leandro Martínez* Department of Physical Chemistry, Institute of Chemistry, University of Campinas, CP 6154-13083-970, Campinas, SP Brazil ABSTRACT: Nuclear hormone receptors (NR) are tran- scription factors that activate gene expression in response to ligands. Structural and functional studies of the ligand binding domains (LBD) of NRs revealed that the dynamics of their C- terminal helix (H12) is fundamental for NR activity. H12 is rigid and facilitates binding of coactivator proteins in the agonist- bound LBD. In the absence of ligand, H12 exhibits increased exibility. To provide a comprehensive picture of the H12 conformational equilibrium, extensive molecular dynamics simulations of the LBD of the PPARγ receptor in the presence or absence of ligand, and of coactivators and corepressor peptides, were performed. Free-energy proles of the conforma- tional variability of the H12 were obtained from more than four microseconds of simulations using adaptive biasing-force calculations. Our results demonstrate that, without ligand, multiple conformations of the H12 are accessible, including agonist-like conformations. We also conrm that extended H12 conformations are not accessible at ordinary temperatures. Ligand binding stabilizes the agonist H12 conformation relative to other structures, promoting a conformational selection. Similar eects are observed with coactivator association. The presence of corepressor peptides stabilizes conformations not allowed in the ligand-free, Rosiglitazone-bound or coactivator-bound LBDs. Corepressor binding, therefore, induces a conformational transition in the protein. Nevertheless, initial stages of corepressor dissociation could be induced by the ligand as it stabilizes the H12 in agonist form. Therefore, the present results provide a comprehensive picture of the H12 motions and their functional implications, with molecular resolution. 1. INTRODUCTION Nuclear hormone receptors (NRs) are transcription factors modulated by ligand binding. Most NRs share the same overall architecture, consisting in three domains: a variable N-terminal domain, a well-preserved DNA binding domain (DBD), and a C-terminal ligand binding domain (LBD). Ligand binding can modulate transcription by controlling the structure and dynamics the LBDs. The LBDs of dierent receptors share similar structures, composed by about 12 helices which are packed in three approximately perpendicular layers, forming an α-helical sandwich. 1,2 In the majority of crystallographic models obtained so far, the ligands are completely buried in the LBDs, such that some dynamical behavior is expected for ligand binding and dissociation, at least. Furthermore, many receptors are able to promote transcription even in the absence of ligand, suggesting that the structures of the LBDs mediating each functional response exist in a dynamical equilibrium, which exists independently but is aected by ligand binding. 3 The rearrangement of the LBDs that is mostly associated with the functional response of the receptors involves its C- terminal helix, the Helix 12 (H12). The conformation of the H12 is determinant for the exposure of interaction sites for coactivator and corepressor proteins. In the presence of agonist ligands, the LBDs recruit coactivator proteins, initiating the signal cascade leading to target gene transcription. On the other side, in the absence of ligand or in the presence of antagonists, the conformation of the H12 favors the interactions of LBDs with corepressor proteins, which suppress the target genes. The conformational variability of the H12 is, therefore, one of the fundamental aspects governing NR function. The dynamic nature of this helix was recognized early with the determination of crystal structure of ligand-bound (holo) and ligand-free (apo) retinoic acid receptors. While the structure of the apo-retinoic X receptor-α (RXRα) 4 displayed an extended H12 not contacting the body of the LBD, the holo-retinoic acid receptor-γ 5 was found with the H12 packed over the LBD in a compact form. This notable dierence suggested originally that the LBDs of NRs would exist in a stable apo conformation in the absence of ligand which would shift to the compact holo structure upon ligand binding. 68 There are two general models for the mechanisms a ligand can aect the function of a biological macromolecule. The rst model, known by conformational selection, describes an intrinsically dynamic protein in the presence or absence of the ligand. The ligand binds to a subset of the conformations, shifting the equilibrium toward this subsets and, thus, Received: October 7, 2015 Revised: November 23, 2015 Published: November 23, 2015 Article pubs.acs.org/JPCB © 2015 American Chemical Society 15418 DOI: 10.1021/acs.jpcb.5b09824 J. Phys. Chem. B 2015, 119, 1541815429

Conformational Diversity of the Helix 12 of the Ligand ...leandro.iqm.unicamp.br/leandro/shtml/publications/2015...Molecular dynamics (MD) simulations further supported the possibility

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

  • Conformational Diversity of the Helix 12 of the Ligand BindingDomain of PPARγ and Functional ImplicationsMariana R. B. Batista and Leandro Martínez*

    Department of Physical Chemistry, Institute of Chemistry, University of Campinas, CP 6154-13083-970, Campinas, SP Brazil

    ABSTRACT: Nuclear hormone receptors (NR) are tran-scription factors that activate gene expression in response toligands. Structural and functional studies of the ligand bindingdomains (LBD) of NRs revealed that the dynamics of their C-terminal helix (H12) is fundamental for NR activity. H12 is rigidand facilitates binding of coactivator proteins in the agonist-bound LBD. In the absence of ligand, H12 exhibits increasedflexibility. To provide a comprehensive picture of the H12conformational equilibrium, extensive molecular dynamicssimulations of the LBD of the PPARγ receptor in the presenceor absence of ligand, and of coactivators and corepressorpeptides, were performed. Free-energy profiles of the conforma-tional variability of the H12 were obtained from more than fourmicroseconds of simulations using adaptive biasing-forcecalculations. Our results demonstrate that, without ligand, multiple conformations of the H12 are accessible, includingagonist-like conformations. We also confirm that extended H12 conformations are not accessible at ordinary temperatures.Ligand binding stabilizes the agonist H12 conformation relative to other structures, promoting a conformational selection.Similar effects are observed with coactivator association. The presence of corepressor peptides stabilizes conformations notallowed in the ligand-free, Rosiglitazone-bound or coactivator-bound LBDs. Corepressor binding, therefore, induces aconformational transition in the protein. Nevertheless, initial stages of corepressor dissociation could be induced by the ligand asit stabilizes the H12 in agonist form. Therefore, the present results provide a comprehensive picture of the H12 motions andtheir functional implications, with molecular resolution.

    1. INTRODUCTIONNuclear hormone receptors (NRs) are transcription factorsmodulated by ligand binding. Most NRs share the same overallarchitecture, consisting in three domains: a variable N-terminaldomain, a well-preserved DNA binding domain (DBD), and aC-terminal ligand binding domain (LBD). Ligand binding canmodulate transcription by controlling the structure anddynamics the LBDs. The LBDs of different receptors sharesimilar structures, composed by about 12 helices which arepacked in three approximately perpendicular layers, forming anα-helical “sandwich”.1,2 In the majority of crystallographicmodels obtained so far, the ligands are completely buried in theLBDs, such that some dynamical behavior is expected for ligandbinding and dissociation, at least. Furthermore, many receptorsare able to promote transcription even in the absence of ligand,suggesting that the structures of the LBDs mediating eachfunctional response exist in a dynamical equilibrium, whichexists independently but is affected by ligand binding.3

    The rearrangement of the LBDs that is mostly associatedwith the functional response of the receptors involves its C-terminal helix, the Helix 12 (H12). The conformation of theH12 is determinant for the exposure of interaction sites forcoactivator and corepressor proteins. In the presence of agonistligands, the LBDs recruit coactivator proteins, initiating thesignal cascade leading to target gene transcription. On the other

    side, in the absence of ligand or in the presence of antagonists,the conformation of the H12 favors the interactions of LBDswith corepressor proteins, which suppress the target genes.The conformational variability of the H12 is, therefore, one

    of the fundamental aspects governing NR function. Thedynamic nature of this helix was recognized early with thedetermination of crystal structure of ligand-bound (holo) andligand-free (apo) retinoic acid receptors. While the structure ofthe apo-retinoic X receptor-α (RXRα)4 displayed an extendedH12 not contacting the body of the LBD, the holo-retinoic acidreceptor-γ5 was found with the H12 packed over the LBD in acompact form. This notable difference suggested originally thatthe LBDs of NRs would exist in a stable apo conformation inthe absence of ligand which would shift to the compact holostructure upon ligand binding.6−8

    There are two general models for the mechanisms a ligandcan affect the function of a biological macromolecule. The firstmodel, known by “conformational selection”, describes anintrinsically dynamic protein in the presence or absence of theligand. The ligand binds to a subset of the conformations,shifting the equilibrium toward this subsets and, thus,

    Received: October 7, 2015Revised: November 23, 2015Published: November 23, 2015

    Article

    pubs.acs.org/JPCB

    © 2015 American Chemical Society 15418 DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    pubs.acs.org/JPCBhttp://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-000.jpg&w=229&h=134

  • modulating the activity. The bound state, therefore, is a subsetof conformations which is stabilized by ligand binding, althoughthese same conformations are populated in the absence of theligand.9 By opposition, the “induced fit” model assumes that thebound state is characterized by conformations which are notsampled in the absence of ligand. Ligand binding promotes aconformational transition from the unbound to the boundstate.9−11 Simplified representations of these models are shownin Figure 1.

    The large conformational shift observed for the H12 in thefirst crystallographic models was consistent with an induced-fitmechanism. The LBD, therefore, should exist in a stable apoform which could suppress gene transcription, and betransformed upon ligand binding to the holo active form.The fundamental conformational change should involve theH12 and the exposure of coactivator or corepressor interactionsurfaces.Nevertheless, the evidence for the induced-fit mechanism

    were not supported by further functional and structuraldata.12,13 For instance, no other apo-LBD models were foundin which the H12 assumed the conformation observed forRXRα, unless when induced by crystallographic packing.14

    Crystallographic apo-LBD models, on the other side, wereobtained in a compact form, similar to holo-RXRγ, for theligand-free Peroxissome-activated activated receptor-γ(PPARγ),15 estrogen receptor-γ (ERγ),16 and for orphanreceptors. Additionally, hydrogen−deuterium exchange experi-ments have shown that the H12 protects the surface of thethyroid hormone receptor (TR) LBD even in the absence ofligand.17 Finally, many receptors, such as PPARγ, display basalactivity; that is, they promote gene transcriptions and thusallow the recruitment of coactivator proteins, also withoutligand.Molecular dynamics (MD) simulations further supported the

    possibility of a more subtle conformational equilibrium of theH12 of NR LBDs. For instance, our group and others haveshown that ligand binding and dissociation routes do not seemto involve major displacements of the H12.12,18−23 Morerecently, we demonstrated that the experimentally observedfluorescence anisotropy decay rates of a probe bound to theH12 of PPARγ are consistent with an H12 which is persistently

    bound to the body of the LBD.24 Therefore, no largeconformational changes should be expected for the H12.24

    Additionally, extended MD simulations performed by Fratev25

    have shown that extended H12 conformations are not favorablefor the estrogen receptor and that the H12 equilibrium betweenagonist and antagonist conformations is subtly affected byreceptor dimerization.Experimental and computational evidence, therefore, is

    suggesting that the activation of the NR LBDs is of theconformational-selection type. The receptors, or morespecifically the LBDs, appear to sample both apo and holoconformations, which might be preferentially stabilized byligand binding. The most promising target for computationalstudies of this conformational equilibrium is the LBD of PPARreceptors. PPAR crystallographic models were obtained in theapo form, in the holo form with coactivator peptides bound(Figure 2A), and in the form bound to both antagonists and

    corepressor peptides (Figure 2B).15,26 Therefore, the conforma-tional variability of the receptor induced, or favored, by each setof ligands and cofactors is relatively well-known.In this work, we aim to describe in details the conformational

    equilibrium of the H12 of PPARγ. We perform extensive MDsimulations with modern free energy calculation methods toobtain a comprehensive profile of the stability of the LBD as afunction of the conformations probed by the H12 in variousconditions. We show that the free energy barriers involved areconsistent with a conformational-selection model for ligand andcoactivator action, and an induced-fit model for corepressorbinding. No extended or highly displaced H12 conformationsare accessible at ordinary temperatures. Therefore, we provide acomprehensive description of the possible movements of themost significant transcription activation switch of PPARγ, thatis probably valid for other NRs.

    ■ MATERIALS AND METHODSMolecular Dynamics simulations were performed for the LBDof PPAR starting from the crystallographic models 2PRG15 or1KKQ,26 for the representation of apo, holo, cofactor free,coactivator-bound, and corepressor-bound PPAR LBD, as willbe discussed.

    Figure 1. Sketch of the two models describing the role of ligands inthe conformational equilibrium of receptors. In the conformationalselection mechanism, the ligand binds a subset of conformations whichare populated even in the absence of ligand, and stabilizes this subsetwhich is identified as the bound state. In the induced fit mechanism,the conformational transition to the bound state occurs after ligandbinding and is promoted by it.

    Figure 2. Crystallographic models for PPARγ. (A) Superposition ofthe LBD of PPARγ in the presence (cyan) or absence (red) of thenatural ligand Rosiglitazone. (B) Crystallographic models of PPARαbound to an antagonist ligand and a corepressor peptide. Thecoactivator and corepressor peptides are colored in blue.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15419

    http://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-001.jpg&w=239&h=130http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-002.jpg&w=239&h=159

  • In all cases, the structures were solvated with Packmol27,28

    with about 22 thousand water molecules, and sodium andchloride ions to render the system neutral. Simulations wereperformed using the CHARMM27 force-field29 for the proteinand peptides, the parameters described by Anders et al.30 forRosiglitazone, and the TIP3P model for water.31 All energyminimization steps and simulations were performed withNAMD.32 Equilibration and simulations were performed inthe NPT ensemble at 298.15K and 1 atm using a time-step of 2fs. Covalent bonds involving hydrogen atoms were kept rigidaccording to standard protocols. Temperature was controlledusing Langevin dynamics with a friction coefficient of 10 ps−1.The Nose-̀Hoover algorithm was used for pressure control,with a piston oscillation period of 200 fs and decay rate of 100fs. Long-range electrostatic interactions were computed withthe PME method. A cutoff of 12 Å was used for van der Waalsinteractions.The systems were equilibrated with (1) 1000 steps of

    conjugate-gradient (CG) energy minimization followed by 200ps MD keeping all protein, ligand and cofactor atoms fixed, forsolvent relaxation. (2) 500 CG minimization steps followed by200 ps MD with fixed Cα coordinates, allowing the side-chainsof the protein to relax. (3) 2 ns MD without any restrictions.The final structure of this last MD was used for productionruns.The Rosiglitazone and coactivator bound crystallographic

    structure of PPARγ (2PRG) was used for the construction ofthe models for (1) apo-PPARγ, for which the ligand and thecoactivator peptide were removed, (2) holo-PPARγ, for whichonly the coactivator peptide was removed, and (3) ligand-free,coactivator bound PPARγ, for which only the ligand wasremoved from the crystallographic model. The PPARα(1KKQ) crystallographic model was used for the simulationof the corepressor-bound LBD, so the antagonist ligand wasremoved.VMD was used for visualization and figure preparation.33

    Cpptraj from the AmberTools 14 suite was used for RMSD-based conformational clustering of the structures by means of ahierarchical algorithm.34,35 The most representative structure ofeach cluster was obtained by Cpptraj and used to producefigures. Different RMSD criteria for the definition of theclusters were tested, and the one that provided the most cleardiscrimination of clusters was considered.Adaptive Biasing-Force Simulations. All simulations

    performed here aimed the comprehensive profiling of the freeenergies involved in conformational transitions of the H12 ofthe LBD. The adaptive biasing-force (ABF) algorithm waschosen for mapping the free energies along reactioncoordinates involving displacements of the H12 of the LBDunder different conditions. The ABF method is currently one ofthe most efficient strategies for accelerated conformationalsampling in MD simulations.36,37

    Briefly, ABF is a strategy for accelerated sampling and freeenergy profiling along a reaction coordinate. Given a reactioncoordinate ξ that can be defined in terms of the coordinates ofthe atoms of the system, the ensemble average force acting onthe atoms at ξ provides the gradient of the free energy.37

    Therefore, if an effective sampling of conformations isperformed within the range of ξ of interest, the free energyprofile relative to any reference state can be reconstructed fromthe average force.The ABF method consists on performing MD simulations in

    which the average force acting on atoms that define the reaction

    coordinate of interest is computed. Once the average force oneach atom is adequately estimated for a given reactioncoordinate ξ, a force with same modulus but with oppositedirection is added to the system whenever it samples that samereaction coordinate. From then on, the forces acting along thereaction coordinate at that point will be on average null, andthe system will move diffusively. When a diffusive motion alongthe complete reaction coordinate of interest is obtained, thefree energy profile along it can be obtained from the biasingforce which was introduced. More detailed descriptions of themethod with rigorous justifications can be obtained else-where.36−40

    In ABF simulations, therefore, it is fundamental to define anadequate reaction coordinate. Here, we aimed to study theconformational variability of the H12 of the LBD of PPARreceptors, therefore the reaction coordinate should representdisplacements of this helix. The RMSD of the helix relative tosome conformation of choice would be, therefore, the naturalchoice. However, many different conformations might displaythe same RMSD relative to a single reference structure, suchthat the free energy computed for that RMSD would not berepresentative of a well-defined subset of conformations. Thisproblem is greatly reduced if multiple reference coordinates areused.40 That is, if the reaction coordinate is defined as theRMSDs relative not to one, but to multiple structures. Thereaction coordinate then becomes multidimensional. Acomplication arises, then, on the definition of the multiplereference coordinates.Here, ABF simulations were used first as an auxiliary

    enhanced sampling method to obtain different referencecoordinates, than as the tool to map the multidimensionalfree energy profile. The first set of simulations consisted in ABFsimulations in which a single reference structure was used, i.e.,the H12 conformation present in the crystallographic model.The conformational variability of the H12 in this simulationwas used to obtain an alternative conformation displaying thelargest displacements from the crystallographic model, butpreserving the secondary structure of the helix. Using this newH12 conformation and the conformation observed in thecrystallographic structure, the multidimensional free energyprofiles were obtained.The ABF simulations were performed as implemented in

    NAMD.32 In the auxiliary simulations performed for mappingH12 conformations, the reaction coordinate was defined as theRMSD relative to the agonist H12 conformation of 2PRG. TheABF force was applied only on the loop connecting helices 11and 12, because its application on all residues of the H12promoted a rapid disruption of the secondary structure insteadof the effective displacement of the helix. The RMSD of thisloop was sampled by ABF simulations within 1 and 10 Å, with aprecision of 0.1 Å. The movements of the atoms outside theregion of interest were avoided by the use of an harmonicboundary potential with a 10 kcal mol−1 Å2 force constant. Fourindependent 80 ns ABF simulations were performed with thisprotocol, and as mentioned above the H12 of largestdisplacement, but preserving the secondary structure, wasselected, for the productive ABF simulations that followed. Theagonist H12 conformation and the alternative conformationselected are represented in Figure 3.Production ABF runs used the two reference structures to

    define a multidimensional reaction coordinate based on theRMSDs of the H12 relative to both models. The two modelswill be referred as the “agonist conformation”, obtained from

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15420

    http://dx.doi.org/10.1021/acs.jpcb.5b09824

  • the holo-PPARγ crystallographic structure, and the “alternativeconformation”, obtained from the auxiliary ABF run (Figure 3).The RMSDs defining the reaction coordinates were computedfor Cα atoms of residues 449 to 477, containing the C-terminalend of the H11, the loop between H11 and H12, and the H12.The same range of RMSDs, precision and boundary potentialsof auxiliary ABF run were used. However, the sampling herewas exhaustive. The minimum sampling time for the force ateach reaction coordinate was 1 ps.For the apo-PPARγ and the Rosiglitazone-bound PPARγ

    models, a total of 1.2 μs of ABF simulations were performed toguarantee convergence of the free energy profiles. Thesesimulations were divided in 12 sets of 100 ns each. For thecorepressor-bound and coactivator-bound models, 1.0 μs ofABF simulations was performed. The convergence of the freeenergy profiles was probed by computing the average deviationof the free energies predicted at the regions of low free energy.The profiles were considered to be converged if the deviation ofthe free energy resulting from the removal of any set of 100 nssimulation was in average smaller than 0.5 kcal mol−1 in allregions differing from the free energy minimum by less than 4kcal mol−1.

    ■ RESULTS AND DISCUSSIONAccessible Conformations of PPARγ H12. The free

    energy profiles of ligand-free and ligand-bound PPARγ areshown in Figure 4. The first noticeable difference between thefree energy surfaces of parts A and B of Figure 4 is that in theligand-free model there is a larger region of low free energies,and two clearly discernible local minimizers. The free energy

    surface of the ligand bound model (Figure 4B), on the otherside, displays a much more narrow low-energy well. At the sametime, the global minimizers of the free energy have similarpositions in both plots.The large well in which the global minimizer of the ligand-

    free receptor is found is composed by many local minimizersthat differ in energy by less than ∼4 kcal mol−1. At 298.15 K,each increase of ∼1 kcal mol−1 in the free energy represents adecreased 80% probability of observation of the higher energystate. Therefore, ∼ 4 kcal mol−1 represents an upper limit theobservable energy states at room temperature, with about 0.1%probability. States differing in energy in less than ∼3 or ∼4 kcalmol−1 from the state that globally minimizes the free energy aresignificantly populated at room temperature. Therefore, theblue and cyan regions of lower energy observed in the freeenergy surfaces of Figure 4 are accessible at room temperature,and illustrate the range of conformations that the H12 canassume. The wider free energy well of the ligand-free PPARγmodel clearly indicates a greater conformational variabilityrelative to the ligand-bound model. This was expected andconfirms the greater flexibility of H12 in the absence ofligands.41−44 The present simulations allow the actual structuralcharacterization of the accessible conformations.The global free energy minima for both ligand-free and

    Rosiglitazone-bound simulations was found close to theRMSD1 = 3 Å and RMSD2 = 4 Å reaction coordinate. Theconformations in the proximities (±0.1 Å) of these regionswere clustered in each system, and the most populatedrepresentative structures were identified. These structures areshown in Figure 5A and B, superposed to the crystallographicholo-PPARγ model. The most populated position of the H12on both ligand-free or ligand-bound free energy minima differsonly slightly from the H12 observed in crystallographic modelsof the ligand-bound PPARγ. In this conformation, the H12 isperpendicular to H3, and attached to the body of the LBD by anetwork of hydrogen bonds which is similar to the oneobserved for the ligand-bound structure (Figure 5C). Theseinteractions involve residues Glu324 (from the loop betweenH4 and H5), Arg397 (loop within H8 and H9), Arg443 (fromH11), and Tyr477 (from H12).45 Interestingly, it was alreadydemonstrated that the mutation of these residues to Alaninereduces significantly the basal activity of PPARγ, thussupporting their stabilization of the active H12 conformationeven in the absence of agonist ligands.46,47

    Figure 3. Agonist and alternative conformations used as referencestructures in the multidimensional ABF simulations.

    Figure 4. Free energy surfaces of the conformational variability of the H12 of PPARγ. (A) Ligand-free LBD. (B) Rosiglitazone-bound LBD. Thewider well of the ligand-free free energy surfaces implies a greater conformational variability.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15421

    http://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-003.jpg&w=143&h=119http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-004.jpg&w=425&h=149

  • Naturally, the basal activity depends on the association of acoactivator protein. Indeed, the position of the most relevantresidues responsible for the establishment of the coactivatorbinding surface are also preserved in the ligand-free global freeenergy minimum. For instance, the interactions of thecoactivator peptide SRC-1 depend on the relative positions ofresidues Lys301 (from H3) and Glu471 (from H12), and themutation of these two residues to alanine can decrease in about30% the basal activity of the receptor.46

    Therefore, our simulations indicate that the basal activity ofPPARγ can be consistently associated with LBD conformationswhich are similar to the ones assumed by the receptor upon

    ligand binding. Furthermore, these conformations are not onlyaccessible, but even in the absence of ligand they comprise thefree energy minimum and, therefore, remain as the mostpopulated set of conformers of the H12. This observationjustifies the fact that, particularly for PPARs, structures in activeform were obtained with a great variety of ligands and evenwithout ligands,15,48−50 while different H12 conformations wereexperimentally obtained only when the H12 is displaced bycorepressor binding proteins.26

    However, without ligand, the active conformation of H12competes with multiple conformations that have slightly higherenergies and are accessible at ordinary temperatures. Thus, the

    Figure 5. Comparison of the representative conformations of the free energy minima (cyan) and the crystallographic model of holo-PPARγ15 (red)for (A) ligand-free PPARγ simulations and (B) Rosiglitazone-bound PPARγ simulations. (C) Detailed view of the interactions that stabilizes H12 inactive conformation.

    Figure 6. Most representative conformations of the H12 obtained by clustering analysis of the ligand-free PPARγ ABF simulation. In all figures, thecrystallographic model of holo-PPARγ is depicted in red. (A) Representative structure of the most populated cluster. (B) Representative structure ofthe second most important cluster. (C) Representative structure of the third and minor cluster. (D) Representation of the conformational variabilityexperienced by the H12 in cluster 1. The conformation of the global free energy minimizer is depicted in blue. (E) Acessible conformations selectedto clustering analysis: the red, green, and blue points indicate the confomations of clusters 1, 2, and 3, respectively.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15422

    http://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-005.jpg&w=370&h=113http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-006.jpg&w=303&h=327

  • conformational variability of the H12 in the ligand-free LBDwas studied by clustering the structures sampled by the ABFsimulation. Conformations which displayed free energies notgreater than 4 kcal mol−1 than the global minimizer wereconsidered to be accessible at room temperature. Theclustering was performed using pairwise RMSDs as thesimilarity measure. The conformations can be clearlydiscriminated in three clusters. Two of the clusters respondedto roughly 97% of the conformations. The mean free energydifference between the two most populated cluster is of only∼0.2 kcal mol−1. A third minor set of structures was classifiedseparately. The most representative conformations of eachcluster are shown in Figure 6A−C, aligned to the holo-PPARγcrystallographic model. Figure 6D represents the conforma-tional variability of the most significant cluster, and Figure 6Eshows the distribution of the conformations of each cluster onthe free energy surface. It is important to remark that theRMSDs resulting from clustering analysis are not necessarilythe same as the ones that define the reaction coordinate, so thatthere is overlap on the distributions of the structures of eachcluster on the free energy surface (Figure 6E).The most representative structure of cluster 1 (Figure 6A-

    cyan) shows that it groups H12 conformations which aresimilar to that of the crystallographic holo-PPARγ (Figure 6A-red). The most populated cluster, therefore, contains the globalfree energy minimum, and is characterized by local fluctuationsof the H12 around the agonist conformation, which allows therecruitment of coactivators (Figure 6D).A representative structure of the second most populated

    cluster of conformations is shown in Figure 6B. Thisconformation will be referred to as the C2 conformationfrom now on. The H12 is clearly displaced from the agonistconformation, being almost parallel to H3. In the C2conformation, the H12 blocks the access to hydrophobicresidues Thr297, Leu311, Gln 314, Val315, Leu318 and Leu468and to the hydrogen bonding Glu471, that form the coactivatorbinding surface. Therefore, this H12 blocks coactivatorrecruitment and is, consequently, inactive.This conformation is remarkably similar to the crystallo-

    graphic structure obtained by Nolte et al. (PDB: 1PRG15) forchain B of the PPARγ without ligand, as shown in Figure 7A.Therefore, the most important conformations of H12 observedin the ABF simulations are consistent with the experimentalH12 conformations. The present results show that, in theabsence of ligand, these conformations are significantlypopulated, and are separated from the minimum free energystructure by ∼2.5 kcal mol−1. Thus, our results support thestructure of Nolte et al. as a representative structure of apo-

    PPARγ in solution, and not simply a crystallographic artifact. Atleast, it is a conformation which might have been stabilized bycrystal contacts, not induced by them, thus possibly beingfunctionally relevant. The C2 conformation is also half wayfrom the agonist to the antagonist conformation observed forcorepressor-bound PPARγ, such that transitions from thisconformation to one or other state can be thought to befacilitated upon binding of agonists, antagonists, or cofactorproteins (Figure 7B).The binding site accessibility is different depending on the

    conformation of the H12. When H12 assumes the agonistconformation, it blocks access to the binding pocket, as shownin Figure 8A. On the other side, the C2 conformation creates

    an aperture at the surface of the binding cavity which mightpermit the access of the ligand, as shown in Figure 8B.Therefore, the C2 conformation might be important for ligandbinding and dissociation. Displacements of the H12 observed inMD simulations of ligand dissociation from the LBD of otherreceptors are consistent with movements of this magni-tude.18−21

    Finally, the representative conformation of cluster 3 (C3conformation) displays the H12 far from the agonistconformation, exposing the binding pocket (Figure 6C). Thisconformation is less populated than others, but is alsoaccessible, and it could be important for facilitating ligandbinding or dissociation. It could also be stabilized by ligands inintermediate association or dissociation stages. It is not,however, consistent with the binding of coactivators norcorepressors, such that it might not have any functionality otherthan providing accessible intermediates for structural transitionsbetween functional states.

    Rosiglitazone Stabilizes the Agonist H12 Conforma-tion. ABF simulations of PPARγ without ligand revealed the

    Figure 7. (A) Superposition of the crystallographic model obtained by Nolte et al.15 (red) and of the representative structure of the second mostimportant cluster (cyan). (B) Superposition of the agonist crystallographic model (red), antagonist crystallographic model (orange), and the C2conformation (cyan).

    Figure 8. Binding site cavity of (A) agonist and (B) C2 conformation.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15423

    http://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-007.jpg&w=359&h=116http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-008.jpg&w=239&h=116

  • existence of a set of conformations of the H12 which areaccessible at ordinary temperatures. These conformations canbe clustered into three sets, where the two most populatedclusters comprise about 97% of the conformations observed. Arepresentative structure of the most populated conformer issimilar to the agonist H12 conformation. The second mostpopulated conformer is displaced from the agonist conforma-tion, and is similar to one of the H12 conformations observedin chain B of apo-PPARγ crystallographic models.15

    When Rosiglitazone, an agonist ligand, is present, the freeenergy surface changes significantly, as can be observed inFigure 4. It is clear from Figure 4B that with Rosiglitazonebound, the global free energy minimum is found in a narrowwell with much lower free energy than any surroundingconformation. The ligand, therefore, stabilizes significantly theglobal free energy minimum relative to other conformations,and only structures which belong to the same free energy wellcan be significantly populated at room temperature.

    The global free energy minimizer is, not surprisingly, verysimilar to the agonist H12 conformation observed in crystallo-graphic models of PPARγ bound to various agonist ligands, asshown in Figure 5B. It is also similar, therefore, to the globalfree energy minimizer of the ligand-free ABF simulation.However, only small perturbations are thermodynamicallyallowed in the ligand-bound state. Clustering of the accessibleconformations in this simulation results in a single dominantcluster comprising essentially all the structures, for which theagonist H12 conformation is a good representative.These results are, as a whole, quite consistent with the

    qualitative picture that has emerged for the flexibility of H12 insolution, from experimental data: In the absence of ligand, theH12 is flexible, and able to assume the agonist conformation,conferring basal activity to PPARγ. However, we demonstrate,here, that the displacements of the H12 are relatively small,particularly if compared to the detachment of the H12 whichwas observed for the apo-RXR crystallographic model.4 This

    Figure 9. (A) Free energy surface of the conformational equilibrium of H12 of PPARγ in the absence of ligand, but bound to a coactivator peptide.(B) Comparison of the representative conformations obtained via ABF simulation (cyan and blue) and the experimental structures (red and orange).

    Figure 10. Representative conformation of each local minimum observed in coactivator-bound ABF simulations. The structures exhibit the H12 in aconformation corresponding to the agonist conformation, which is greatly stabilized by coactivator binding. Crystallographic models (red andorange) and simulation models (cyan and blue) are shown.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15424

    http://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-009.jpg&w=319&h=150http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-010.jpg&w=341&h=238

  • confirms the interpretation of time-resolved fluorescenceanisotropy decay experiments provided in previous works.24,41

    In the presence of agonists, on the other side, the flexibility isgreatly reduced, and the agonist conformation becomes theonly significantly populated one. The ligand, therefore,performs a conformational selection within a range ofconfigurations of the H12 which are accessible in the apo-receptor. The stabilization of the agonist conformation, ofcourse, facilitates the recruitment of coactivator proteins.The Role of Coactivators in H12 Conformational

    Equilibrium. In order to evaluate the role of coactivatorproteins in the conformational equilibrium of the H12, ABFsimulations in the presence of a coactivator peptide SRC-115

    were performed. The free energy surface obtained isrepresented in Figure 9A.The representative structures of accessible conformations of

    the H12 in the presence of the coactivator peptide are alsoagonist-like, as shown in Figure 9B (the dominant clustercomprises more than 99% of the structures). The stabilizationof this conformation promoted by the coactivator is, however,stronger than that promoted by the ligand. The H12, in theminimum free energy conformation, forms the same

    interactions which are present in the previous simulations,but is further stabilized by direct interactions with thecoactivator. Displacements from this conformation, with highenergy, were observed in the ABF simulations, and areconsistent with movements leading to the antagonist crystallo-graphic PPARα H12 conformation. However, these displace-ments require the concerted dragging of the coactivator peptideand are unfavorable. The ligand and the coactivator, therefore,have complementary roles in the stabilization of the agonistH12 conformations.Three free energy minima were observed in the presence of

    the coactivator peptide. However, the actual structuresassociated with these minima exhibit the H12 in conformationswhich are all very similar to the agonist conformation. Onlysmall perturbations in the position of the loop connectinghelices 11 and 12 are observed (Figure 10).Therefore, the association of the coactivator has a strong

    stabilizing effect on the agonist conformation of the H12, whichis more important than the stabilization promoted byRosiglitazone alone. This suggests that the dynamicalequilibriums of ligand binding, H12 rearrangement, andcoactivator recruitment occur at different time-scales. The

    Figure 11. (A) Free energy surface of the conformational equilibrium of H12 of PPARα in the absence of ligand, but bound to a corepressor peptide.(B) Comparison of the representative structure of the minimum energy region (cyan and blue) and the antagonist crystallographic model (red andorange).

    Figure 12. (A) Conformations of H12 similar to the agonist PPARγ model are observed (conformation B) with free energies not greater than ∼2kcal mol−1 than that of the global minimizer These conformations are associated with the displacement of the coactivator peptide, and suggest howthe ligand actively promotes the dissociation of the corepressor. (B) Hydrogen bond (black dashed line) between SMRT CoR and PPARα.Crystallographic models (orange) and simulation models (blue) are shown.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15425

    http://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-011.jpg&w=349&h=144http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-012.jpg&w=418&h=189

  • most plausible scenario would involve ligand binding anddissociation from a coactivator-free receptor. With agonistbinding, agonist H12 conformations are selected, favoringcoactivator recruitment. With coactivator recruitment, this LBDconformation is further stabilized. The dissociation of thecoactivator, then, becomes the limiting step for the restorationof the inactive receptor.The Role of Corepressors in the Conformational

    Equilibrium of Antagonist H12 Conformations. Exper-imentally, the association of corepressors is associated with asignificant displacement of the H12. In order to improve theoverall picture of free energy profiles, we have also performedABF calculations in the presence of the corepressor peptideSMRT.26 This is particularly interesting because configurations

    of the H12 similar to those observed in corepressor-boundcrystallographic models were observed in ABF simulations ofligand-free, ligand-bound, and coactivator-bound PPARγ LBDs,but with energies ∼30 kcal mol−1 greater than that of theagonist conformation, implying that they were not accessible.Therefore, the presence of the corepressor peptide mustpromote a large perturbation of the free energy surface to allowthe antagonist H12 conformations to be sampled.Simulations of the LBD bound to a corepressor were

    performed for PPARα, for which a crystallographic model isavailable.26 The free energy surface obtained is shown in Figure11A. The position of the global minimizer is shifted, and theregions of minimum energy of ligand-free and Rosiglitazone-bound PPARγ become prohibited, while the regions of high

    Figure 13. Sketch of the free energy profile and conformational equilibrium of H12: (A) In the absence of ligand, multiple conformations of H12 areaccessible, with the agonist one being the most stable. (B) Ligand binding promotes a conformational selection and the agonist conformation isstabilized relative to other conformations. The association of the coactivator has a similar and probably additive effect. (C) Corepressor binding, onthe other side, promotes a conformational transition of the H12 that cannot be explained by the accessible conformations of the ligand-free receptor.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15426

    http://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showImage?doi=10.1021/acs.jpcb.5b09824&iName=master.img-013.jpg&w=400&h=475

  • energy sampled in previous runs are stabilized. Therefore, thecorepressor stabilized conformations of the H12 which are notaccessible in its absence, and promotes a relative increase in thefree energies of the agonist H12 conformations.There are overlaps between the subsets of conformation

    accessible in the presence of corepressor and those accessible inthe absence of ligand. Structures very similar to C2conformation are observed in corepressor-bound simulations(Figure 11B). This result suggests that the C2 conformationsupports the interaction with the corepressor peptide. However,the antagonist conformation which is observed with thecorepressor must still be stabilized by antagonist binding.In the corepressor-bound system, the accessible free energy

    surface is very wide, and shows two local minima (Figure 12).One of the local minima corresponds to H12 conformationswhich are similar to the agonist conformations. Interestingly,this agonist, low-energy conformations, required the displace-ment of the corepressor peptide from the crystallographicposition, as shown in Figure 12. In the crystallographic model,the SMRT peptide adopts a three-turn α-helix and docks into ahydrophobic groove formed by helices 3, 4, and 5. Besides that,the corepresor helix is anchored to PPARα by hydrogen bondsbetween the Lys292 (from H3) and the CoR peptide, as shownin Figure 12B.26 The rearrangement of H12 to restore theactive conformation require dragging the corepressor peptide,notably with the disruption of its interaction with Lys292. Theinitial stages of this displacement are not thermodynamicallyprohibited, such that the LBD can assume a full agonistconformation in the presence of a partially displacedcorepressor. Agonist binding, by stabilizing the H12 in theagonist structure, might therefore actively promote thedissociation of the corepressor.Classification of the Conformational Equilibrium of

    the H12. The free energy profiles obtained here allow us toclassify the conformational equilibria induced by the associationof ligands and cofactors according to general models. Figure 13summarizes our observations.In the ligand-free receptor, multiple conformations of the

    H12 are accessible. The most stable conformation is the agonistconformation, but the helix is mobile. Not only multipleminima with small free energy differences are present, but thetransitions between these minima are also of low energy.Indeed, the low free energy well is more clearly described as arugged free energy landscape, spanning various conformations.A simplified free energy plot illustrating the multiple accessiblestates of the ligand-free receptor is shown in Figure 13A.Ligand binding promotes a conformational selection. The

    agonist conformation of the H12 is stabilized relative to otherconformations, as shown in Figure 13B. The ligand does notchange the structural nature of the most stable conformationbut makes of it the only accessible conformation at roomtemperature. Coactivator binding has a similar effect as that ofligand binding, and both effects might be additive. The agonistconformation is stabilized relative to other states, but thestabilization is more pronounced with the coactivator peptide.Therefore, it can also be classified as a conformational selectionmechanism. Ligand or coactivator association causes therelative stabilization of the agonist conformation, withoutinducing conformational changes to the LBD to previouslyunaccessible states.The association of the corepressor, on the other side,

    promotes a conformational transition of the H12 that cannot beanticipated from the accessible conformations of the ligand-free

    receptor. The favorable conformations in the presence of thecorepressor are tenths of kcal mol−1 far from the stablecorepressor-free conformers in the other free energy profiles.Therefore, the H12 can only assume these conformationsinduced by the association of the corepressor. The accessibleconformations are also mutually exclusive, such that the onesthat are accessible without the corepressor become prohibitedwhen it is bound to the LBD. Nevertheless, small displacementsof the corepressor with small energetic cost permit the H12 toassume conformations which are close to the agonist one. Thesimplified representation of the perturbation of the free energysurface introduced by the association of the corepressor isshown in Figure 13C. As the conformations assumed by theH12 in this case can only be observed after the association ofthe corepressor, the mechanism of repression is an induced-fitmechanism.

    ■ CONCLUSIONSWe have performed adaptive biasing-force molecular dynamicssimulations to map the free energy profile of the conforma-tional equilibrium of the Helix 12 of the PPARγ receptor. Thisconformational equilibrium is the most important factorcontrolling transcription activation or repression by NRs.The free energy surfaces obtained allowed us to propose

    general models for ligand and cofactor related conformationalchanges on the receptor. Without ligand, the H12 is able toshift between various conformations, consistently with theexperimental data that demonstrates its increased mobility.These conformations are, at the same time, always consistentwith a compact receptor, thus no detachment of the H12 fromthe body of the receptor is expected. Furthermore, even in theabsence of ligand, the most stable H12 conformation is theagonist conformation, providing a simple explanation for thebasal activity of PPARγ. The accessibility of agonist H12conformations in the absence of ligand was also recentlydemonstrated for the estrogen receptor,25 supporting thegenerality of the results for the other members of the NRsuperfamily.Ligand binding promotes a conformational selection. It

    stabilizes significantly the agonist H12 conformation relative toall other conformations which are accessible in the ligand-freereceptor. A similar effect on the free energy surface is promotedby binding a coactivator peptide to the LBD. In both cases, thefree energy well containing the agonist H12 is narrowed, andonly conformations similar to that observed in the crystallo-graphic model become accessible at room temperature. Thestabilization of the agonist conformation by ligand bindingexposes the coactivator binding surface permanently. Thecoactivator itself stabilizes the same conformation, such that ithas an additive role for the stability of the active form of theLBD.Corepressor binding, on the other side, perturbs completely

    the free energy surface. Conformations which are accessible inthe ligand-free, in the presence of Rosiglitazone, or in thepresence of the coactivator, become unfavorable. The favoredconformations in the presence of the corepressor are those witha large displacement of the H12, and these are not accessible inother conditions. Therefore, the association of the corepressorinduces a conformational transition in the protein, and themechanism of repression follows an induced-fit model.Interestingly, however, the H12 can assume agonist-likeconformations by displacing the corepressor peptide with asmall energetic cost. As the ligand stabilizes the agonist

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15427

    http://dx.doi.org/10.1021/acs.jpcb.5b09824

  • conformation, this might be the mechanism for the initialagonist-induced dissociation of the corepressor.Therefore, in conjunction with previous experimental and

    simulation data, this work contributes to a definitive model ofthe conformational equilibrium of the most important switchfor transcription activation of nuclear hormone receptors.

    ■ AUTHOR INFORMATIONCorresponding Author*(L.M.) Telephone: +55 19 3521 3107. E-Mail: [email protected].

    NotesThe authors declare no competing financial interest.

    ■ ACKNOWLEDGMENTSThe authors thank Fapesp and the Center for ComputationalEngineering and Sciences (CCES) (Grants 2010/16947-9,2011/51348-1, 2013/05475-7, and 2013/09465-6), CNPq(Grants 304058/2013-0 and 470374/2013-6), and Faepex(Grant 596/13) for financial support.

    ■ REFERENCES(1) Kumar, R.; Thompson, E. R. The Structure of the NuclearHormone Receptors. Steroids 1999, 64, 310−319.(2) Aranda, A.; Pascual, A. Nuclear Hormone Receptor and GeneExpression. Physiol. Rev. 2001, 81, 1269−1304.(3) Mackinnon, J. A. G.; Gallastegui, N.; Osguthorpe, D. J.; Hagler,A. T.; Estebanez-Perpina, E. E. Allosteric Mechanisms of NuclearReceptors: Insights from Computational Simulations. Mol. Cell.Endocrinol. 2014, 393, 75−82.(4) Bourguet, W.; Ruff, M.; Chambon, P.; Gronemeyer, H.; Moras,D. Crystal Structure of the Ligand-Binding Domain of the HumanNuclear Receptor RXRα. Nature 1995, 375, 377−382.(5) Renaud, J. P.; Rochel, N.; Ruff, M.; Vivat, V.; Chambon, P.;Gronemeyer, H.; Moras, D. Crystal Structure of the RARγ Ligand-Binding Domain Bound to All Trans Retinoic. Nature 1995, 378,681−689.(6) Moras, D.; Gronemeyer, H. The Nuclear Receptor Ligand-Binding Domain: Structure and Function. Curr. Opin. Cell Biol. 1998,10, 384−391.(7) Gronemeyer, H.; Gustafsson, J. A.; Laudet, V. Principles for theModulation of the Nuclear Receptor Superfamily. Nat. Rev. DrugDiscovery 2004, 3, 950−964.(8) Nettles, K. W.; Greene, G. L. Ligand Control of CoregulatorRecruiment to Nuclear Receptors. Annu. Rev. Physiol. 2005, 67, 309−333.(9) Wang, J.; Shao, Q.; Xu, Z.; Liu, Y.; Yang, Z.; Cossins, B. P.; Jiang,H.; Chen, K.; Shi, J.; Zhu, W. Exploring Transition Pathway and Free-Energy Profile of Large-Scale Protein Conformational Change byCombining Normal Mode Analysis and Umbrella Sampling MolecularDynamics. J. Phys. Chem. B 2014, 118, 134−143.(10) Koshland, D. E. Application of a Theory of Enzyme Specificityto Protein Synthesis. Proc. Natl. Acad. Sci. U. S. A. 1958, 44, 98−104.(11) Bucher, D.; Grant, B. J.; Mccammon, J. A. Induced Fit orConformational Selection? The Role of the Semi-Closed State in theMaltose Binding Protein. Biochemistry 2011, 50, 10530−10539.(12) Martínez, L.; Sonoda, M. T.; Webb, P.; Baxter, J. D.; Skaf, M. S.;Polikarpov, I. Molecular Dynamics Simulations Reveal MultiplePathways of Ligand Dissociation from Thyroid Hormone Receptors.Biophys. J. 2005, 89, 2011−2023.(13) Rastinejad, F.; Ollendorff, V.; Polikarpov, I. Nuclear ReceptorFull-Length Architectures: Confronting Myth and Illusion With highResolution. Trends Biochem. Sci. 2015, 40, 16−24.(14) Tanenbaum, D. M.; Wang, Y.; Williams, S. P.; Sigler, P. B.Crystallographic Comparison of the Estrogen and Progesterone

    Receptor’s Ligand Binding Domains. Proc. Natl. Acad. Sci. U. S. A.1998, 95, 5998−6003.(15) Nolte, R. T.; Wisely, G. B.; Westin, S.; Cobb, J. E.; Lambert, M.H.; Kurokawa, R.; Rosenfeld, M. G.; Willson, T. M.; Glass, C. K.;Milburn, M. V. Ligand Binding and Co-activator Assembly of thePeroxisome Proliferator-Actived Receptor-γ. Nature 1998, 395, 137−143.(16) Wang, Y.; Chirgadze, N. Y.; Briggs, S. L.; Khan, S.; Jensen, E. V.;Burris, T. P. A Second Binding Site for the Hydroxytamoxifen Withinthe Coactivator-Binding Groove of Estrogen Receptor β. Proc. Natl.Acad. Sci. U. S. A. 2006, 103, 9908−9911.(17) Figueira, A. C. M.; Saidemberg, D. M.; Souza, P. C. T.;Martínez, L.; Scanlan, T. S.; Baxter, J. D.; Skaf, M. S.; Palma, M. S.;Webb, P.; Polikarpov, I. Analysis of Agonist and Antagonist Effects onThyroid Hormone Receptor Conformation by Hydrogen/DeuteriumExchange. Mol. Endocrinol. 2011, 25, 15−31.(18) Blondel, A.; Renaud, J. P.; Fischer, S.; Moras, D.; Karplus, M.Retinoic Acid Receptor: A Simulation Analysis of Retinoic AcidBinding and the Resulting Conformational Changes. J. Mol. Biol. 1999,291, 101−115.(19) Kosztin, D.; Izrailev, S.; Schulten, K. Unbinding of Retinoic AcidFrom Its Receptor Studied by Steered Molecular Dynamics. Biophys. J.1999, 76, 188−197.(20) Martínez, L.; Webb, P.; Polikarpov, I.; Skaf, M. S. MolecularDynamics Simulations of Ligand Dissociation From ThyroidHormone Receptors: Evidence of the Likeliest Escape Pathway andIts Implications for the Design of Novel Ligands. J. Med. Chem. 2006,49, 23−26.(21) Martínez, L.; Polikarpov, I.; Skaf, M. S. Only Subtle ProteinConformational Adaptations Are Require for Ligand Binding toThyroid Hormone Receptors: Simulations Using a Novel MultipointSteered Molecular Dynamics Approach. J. Phys. Chem. B 2008, 112,10741−10751.(22) Souza, P. C. T.; Barra, G. B.; Velasco, L. F.; Ribeiro, I. C.;Simeoni, L. A.; Togashi, M.; Webb, P.; Neves, F. A. R.; Skaf, M. S.;Martínez, L.; et al. Helix 12 Dynamics and Thyroid HormoneReceptor Activity: Experimental and Molecular Dynamics Studies ofIle280 Mutants. J. Mol. Biol. 2011, 412, 882−893.(23) Fratev, F.; Tsakovska, I.; Al Sharif, M.; Mihaylova, E.; Pajeva, I.Structural and Dynamical Insights into PPARγ Antagonism: In SilicoStudy of the Ligand-Receptor Interactions of Non-CovalentAntagonists. Int. J. Mol. Sci. 2015, 16, 15405−15424.(24) Batista, M. R. B.; Martínez, L. Dynamics of Nuclear ReceptorHelix-12 Switch of Transcription Activation by Modeling Time-Resolved Fluorescence Anisotropy Decays. Biophys. J. 2013, 105,1670−1680.(25) Fratev, F. Helix 12 Orientation in Estrogen Receptors isMediated by Receptor Dimerization: Evidence from MolecularDynamics Simulations. Phys. Chem. Chem. Phys. 2015, 17, 13403−13420.(26) Xu, H. E.; Stanley, T. B.; Montana, V. G.; Lambert, M. H.;Shearer, B. G.; Cobb, J. E.; McKee, D. D.; Galardi, C. M.; Plunket, K.D.; Nolte, R. T.; et al. Structural Basis for Antagonist-MediatedRecruiment of Nuclear Co-repressors by PPARα. Nature 2002, 415,813−817.(27) Martínez, J. M.; Martínez, L. Packing Optimization forAutomated Generation of Complex System’s Initial Configurationsfor Molecular Dynamics and Docking. J. Comput. Chem. 2003, 24,819−825.(28) Martínez, L.; Andrade, R.; Birgin, E. G.; Martínez, J. M.Packmol: A Package for Building Initial Configurations for MolecularDynamics Simulations. J. Comput. Chem. 2009, 30, 2157−2164.(29) Mackerell, A. D.; Bashford, D.; Bellott, M.; Dunbrack, R. L.;Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; et al.All Atom Empirical Potential for Molecular Modeling and DynamicsStrudies of Protein. J. Phys. Chem. B 1998, 102, 3586−3616.(30) Hansson, A.; Souza, P. C. T.; Silveira, R. L.; Martínez, L.; Skaf,M. S. Charmm Force Field Parametrization of Rosiglitazone. Int. J.Quantum Chem. 2011, 111, 1346−1354.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15428

    mailto:[email protected]:[email protected]://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showLinks?pmid=7760929&crossref=10.1038%2F375377a0&coi=1%3ACAS%3A528%3ADyaK2MXmtVehsLc%253Dhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fjp973084f&pmid=24889800&coi=1%3ACAS%3A528%3ADyaK1cXivVOlsb4%253Dhttp://pubs.acs.org/action/showLinks?pmid=21530542&crossref=10.1016%2Fj.jmb.2011.04.014&coi=1%3ACAS%3A528%3ADC%252BC3MXht1GrtrvLhttp://pubs.acs.org/action/showLinks?pmid=10438609&crossref=10.1006%2Fjmbi.1999.2879&coi=1%3ACAS%3A528%3ADyaK1MXkvFKjtLk%253Dhttp://pubs.acs.org/action/showLinks?pmid=7501014&crossref=10.1038%2F378681a0&coi=1%3ACAS%3A528%3ADyaK2MXhtVSnt73Fhttp://pubs.acs.org/action/showLinks?pmid=26184155&crossref=10.3390%2Fijms160715405&coi=1%3ACAS%3A528%3ADC%252BC2MXhsl2qs7nKhttp://pubs.acs.org/action/showLinks?pmid=15980170&crossref=10.1529%2Fbiophysj.105.063818&coi=1%3ACAS%3A528%3ADC%252BD2MXpvF2rsrs%253Dhttp://pubs.acs.org/action/showLinks?crossref=10.1002%2Fqua.22638&coi=1%3ACAS%3A528%3ADC%252BC3MXjvVKitr8%253Dhttp://pubs.acs.org/action/showLinks?crossref=10.1002%2Fqua.22638&coi=1%3ACAS%3A528%3ADC%252BC3MXjvVKitr8%253Dhttp://pubs.acs.org/action/showLinks?pmid=9640540&crossref=10.1016%2FS0955-0674%2898%2980015-X&coi=1%3ACAS%3A528%3ADyaK1cXjvFeltbk%253Dhttp://pubs.acs.org/action/showLinks?pmid=9876133&crossref=10.1016%2FS0006-3495%2899%2977188-2&coi=1%3ACAS%3A528%3ADyaK1MXjsFOisQ%253D%253Dhttp://pubs.acs.org/action/showLinks?pmid=24094408&crossref=10.1016%2Fj.bpj.2013.07.032&coi=1%3ACAS%3A528%3ADC%252BC3sXhsFGqsL%252FEhttp://pubs.acs.org/action/showLinks?pmid=10406480&crossref=10.1016%2FS0039-128X%2899%2900014-8&coi=1%3ACAS%3A528%3ADyaK1MXktFKgsL0%253Dhttp://pubs.acs.org/action/showLinks?pmid=11845213&crossref=10.1038%2F415813a&coi=1%3ACAS%3A528%3ADC%252BD38XhsFersr0%253Dhttp://pubs.acs.org/action/showLinks?pmid=25435400&crossref=10.1016%2Fj.tibs.2014.10.011&coi=1%3ACAS%3A528%3ADC%252BC2cXhvFCnsbnMhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fjm050805n&pmid=16392786&coi=1%3ACAS%3A528%3ADC%252BD2MXht1Kjtr%252FJhttp://pubs.acs.org/action/showLinks?pmid=15520817&crossref=10.1038%2Fnrd1551&coi=1%3ACAS%3A528%3ADC%252BD2cXptFemtrY%253Dhttp://pubs.acs.org/action/showLinks?pmid=15520817&crossref=10.1038%2Fnrd1551&coi=1%3ACAS%3A528%3ADC%252BD2cXptFemtrY%253Dhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fjp4105129&pmid=24350625&coi=1%3ACAS%3A528%3ADC%252BC3sXhvFKhsbnMhttp://pubs.acs.org/action/showLinks?pmid=25927714&crossref=10.1039%2FC5CP00327J&coi=1%3ACAS%3A528%3ADC%252BC2MXms1ajuro%253Dhttp://pubs.acs.org/action/showLinks?pmid=11427696&coi=1%3ACAS%3A528%3ADC%252BD3MXlt1Ohsrs%253Dhttp://pubs.acs.org/action/showLinks?pmid=12692791&crossref=10.1002%2Fjcc.10216&coi=1%3ACAS%3A528%3ADC%252BD3sXjs1ens70%253Dhttp://pubs.acs.org/action/showLinks?pmid=9600906&crossref=10.1073%2Fpnas.95.11.5998&coi=1%3ACAS%3A528%3ADyaK1cXjtlKhurw%253Dhttp://pubs.acs.org/action/showLinks?pmid=16782818&crossref=10.1073%2Fpnas.0510596103&coi=1%3ACAS%3A528%3ADC%252BD28XmvVajtLk%253Dhttp://pubs.acs.org/action/showLinks?pmid=16782818&crossref=10.1073%2Fpnas.0510596103&coi=1%3ACAS%3A528%3ADC%252BD28XmvVajtLk%253Dhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fjp803403c&pmid=18681473&coi=1%3ACAS%3A528%3ADC%252BD1cXptlCjt78%253Dhttp://pubs.acs.org/action/showLinks?pmid=16590179&crossref=10.1073%2Fpnas.44.2.98&coi=1%3ACAS%3A528%3ADyaG1cXms1Khtw%253D%253Dhttp://pubs.acs.org/action/showLinks?pmid=24911885&crossref=10.1016%2Fj.mce.2014.05.017&coi=1%3ACAS%3A528%3ADC%252BC2cXhtlGmtLbOhttp://pubs.acs.org/action/showLinks?pmid=19229944&crossref=10.1002%2Fjcc.21224&coi=1%3ACAS%3A528%3ADC%252BD1MXptleqsb8%253Dhttp://pubs.acs.org/action/showLinks?pmid=24911885&crossref=10.1016%2Fj.mce.2014.05.017&coi=1%3ACAS%3A528%3ADC%252BC2cXhtlGmtLbOhttp://pubs.acs.org/action/showLinks?pmid=15709961&crossref=10.1146%2Fannurev.physiol.66.032802.154710&coi=1%3ACAS%3A528%3ADC%252BD2MXjsFeksr8%253Dhttp://pubs.acs.org/action/showLinks?pmid=21106879&crossref=10.1210%2Fme.2010-0202&coi=1%3ACAS%3A528%3ADC%252BC3MXhvFGrur0%253Dhttp://pubs.acs.org/action/showLinks?pmid=9744270&crossref=10.1038%2F25931&coi=1%3ACAS%3A528%3ADyaK1cXmtVekt7k%253Dhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fbi201481a&pmid=22050600&coi=1%3ACAS%3A528%3ADC%252BC3MXhsVSit7vM

  • (31) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R.W.; Klein, M. L. Comparison of simple potential functions forsimulating liquid water. J. Chem. Phys. 1983, 79, 926−935.(32) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid,E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kale,̀ L.; Schulten, K. ScalableMolecular Dynamics with Namd. J. Comput. Chem. 2005, 26, 1781−1802.(33) Humphrey, W.; Dalke, A.; Schulten, K. Vmd: Visual MolecularDynamics. J. Mol. Graphics 1996, 14, 33−38.(34) Shao, J.; Tanner, S. W.; Thompson, N.; Cheatham, T. E.Clustering Molecular Dynamics Trajectories: 1. Characterizing thePerformance of Different Clustering Algorithms. J. Chem. TheoryComput. 2007, 3, 2312−2334.(35) Arrar, M.; Turnham, R.; Pierce, L.; de Oliveira, C. A. F.;Mccammon, J. A. Structural Insight into the Separate Roles of InositolTetraphosphate and Deacetylase-Activating Domain in Activation ofHistone Deacetylase 3. Protein Sci. 2013, 22, 83−92.(36) Henin, J.; Chipot, C. Overcoming Free Energy Barriers UsingUnconstrained Molecular Dynamics Simulations. J. Chem. Phys. 2004,121, 2904−2914.(37) Chipot, C.; Henin, J. Exploring the Free-Energy Landscape of aShort Peptide Using an Average Force. J. Chem. Phys. 2005, 123,244906−244911.(38) Chipot, T.; Lelievre, T. Enhanced Sampling of MultidimensionalFree-Energy Landscapes Using Adaptive Biasing Forces. SIAM J. Appl.Math. 2011, 71, 1673−1695.(39) Darve, E.; Rodriguez-Gomez, D.; Pohorille, A. Adaptive BiasingForce Method for Scalar and Vector Free Energy. J. Chem. Phys. 2008,128, 144120−144134.(40) Comer, J.; Gumbart, J. C.; Heńin, J.; Lelievre, T.; Pohorille, A.;Chipot, C. The Adaptive Biasing Force Method: Everything YouAlways Wanted To Know but Were Afraid To Ask. J. Phys. Chem. B2015, 119, 1129−1151.(41) Kallenberger, B. C.; Love, J. D.; Chatterjee, V. K. K.; Schwabe, J.W. R. A Dynamic Mechanism of Nuclear Receptor Activation and ItsPerturbation in Human Disease. Nat. Struct. Biol. 2003, 10, 136−140.(42) Nagy, L.; Schwabe, J. Mechanism of the Nuclear ReceptorMolecular Switch. Trends Biochem. Sci. 2004, 29, 317−324.(43) Johnson, B. A.; Wilson, E. M.; Li, Y.; Moller, D. E.; Smith, R. G.;Zhou, G. Ligand-Induced Stabilization of PPARγ Monitored by NMRSpesctroscopic: Implications for Nuclear Receptor Activation. J. Mol.Biol. 2000, 298, 187−194.(44) Hughes, T. S.; Chalmers, M. J.; Novick, S.; Kuruvilla, D. S.;Chang, M. R.; Kamenecka, T. M.; Rance, M.; Johnson, B. A.; Burris, T.P.; Griffin, P. R.; et al. Ligand and Receptor Dynamics Contribute Tothe Mechanism of Graded PPARγ Agonism. Structure 2012, 20, 139−150.(45) Zoete, V.; Grosdidier, A.; Michielin, O. PeroxissomeProliferator-Actived Receptor Structures: Ligand Specificity, MolecularSwitch and Interactions with Regulators. Biochim. Biophys. Acta, Mol.Cell Biol. Lipids 2007, 1771, 915−925.(46) Molnar, F.; Matilainen, M.; Carlberg, C. Structural Determi-nants of the Agonist-Independent Association of Humam PeroxisomeProliferator Activated Receptors with Coactivators. J. Biol. Chem. 2005,280, 26543−26556.(47) Gurnell, M.; Wentworth, J. M.; Agostini, M.; Adams, M.;Collingwood, T. N.; Provenzano, C.; Browne, P. O.; Rajanayagam, O.;Burris, T. P.; Schwabe, J. W.; et al. A Dominant Negative PeroxisomeProliferator-Activated Receptor Gamma Mutant Is a ConstitutiveRepressor and Inhibits PPARγ-Mediated Adipogenesis. J. Biol. Chem.2000, 275, 5754−5759.(48) Bruning, J. B.; Chalmers, M. J.; Prasad, S.; Busby, S. A.;Kamenecka, T. M.; Nettles, K. W.; Griffin, P. R.; He, Y. PartialAgonists Activate PPARγ Using Helix 12 Independent Mechanism.Structure 2007, 15, 1258−1271.(49) Waku, T.; Shiraki, T.; Oyama, T.; Maebara, K.; Nakamori, R.;Morikawa, K. The nuclear Receptor PPARγ Individually Responds ToSerotonin and Fatty Acid-Metabolites. EMBO J. 2010, 29, 3395−3407.

    (50) Montanari, R.; Saccoccia, F.; Scotti, E.; Crestani, M.; Godio, C.;Gilardi, F.; Loiodice, F.; Fracchiolla, G.; Laghezza, A.; Tortorella, P.;et al. Crystal Structure of the Peroxisome Proliferator-ActivatedReceptor γ (PPARγ) Ligand Binding Domain Complexed With aNovel Partial Agonist: A New Region of the Hydrophobic PocketCould Be Exploited For Drug Design. J. Med. Chem. 2008, 51, 7768−7776.

    The Journal of Physical Chemistry B Article

    DOI: 10.1021/acs.jpcb.5b09824J. Phys. Chem. B 2015, 119, 15418−15429

    15429

    http://dx.doi.org/10.1021/acs.jpcb.5b09824http://pubs.acs.org/action/showLinks?pmid=10681562&crossref=10.1074%2Fjbc.275.8.5754&coi=1%3ACAS%3A528%3ADC%252BD3cXhsFKktLc%253Dhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fjp506633n&pmid=25247823&coi=1%3ACAS%3A528%3ADC%252BC2cXhsF2nurvPhttp://pubs.acs.org/action/showLinks?pmid=15291601&crossref=10.1063%2F1.1773132&coi=1%3ACAS%3A528%3ADC%252BD2cXmt1Smurk%253Dhttp://pubs.acs.org/action/showLinks?pmid=12536206&crossref=10.1038%2Fnsb892&coi=1%3ACAS%3A528%3ADC%252BD3sXms1KjtA%253D%253Dhttp://pubs.acs.org/action/showLinks?pmid=17937915&crossref=10.1016%2Fj.str.2007.07.014&coi=1%3ACAS%3A528%3ADC%252BD2sXhtFKhsLzIhttp://pubs.acs.org/action/showLinks?pmid=16396572&crossref=10.1063%2F1.2138694&coi=1%3ACAS%3A528%3ADC%252BD28Xit1Cgtg%253D%253Dhttp://pubs.acs.org/action/showLinks?pmid=15276186&crossref=10.1016%2Fj.tibs.2004.04.006&coi=1%3ACAS%3A528%3ADC%252BD2cXlsFentb4%253Dhttp://pubs.acs.org/action/showLinks?pmid=8744570&crossref=10.1016%2F0263-7855%2896%2900018-5&coi=1%3ACAS%3A528%3ADyaK28Xis12nsrg%253Dhttp://pubs.acs.org/action/showLinks?pmid=8744570&crossref=10.1016%2F0263-7855%2896%2900018-5&coi=1%3ACAS%3A528%3ADyaK28Xis12nsrg%253Dhttp://pubs.acs.org/action/showLinks?pmid=20717101&crossref=10.1038%2Femboj.2010.197&coi=1%3ACAS%3A528%3ADC%252BC3cXhtVejtr3Lhttp://pubs.acs.org/action/showLinks?crossref=10.1137%2F10080600Xhttp://pubs.acs.org/action/showLinks?crossref=10.1063%2F1.445869&coi=1%3ACAS%3A528%3ADyaL3sXksF2htL4%253Dhttp://pubs.acs.org/action/showLinks?crossref=10.1137%2F10080600Xhttp://pubs.acs.org/action/showLinks?pmid=10764590&crossref=10.1006%2Fjmbi.2000.3636&coi=1%3ACAS%3A528%3ADC%252BD3cXitl2ru78%253Dhttp://pubs.acs.org/action/showLinks?pmid=10764590&crossref=10.1006%2Fjmbi.2000.3636&coi=1%3ACAS%3A528%3ADC%252BD3cXitl2ru78%253Dhttp://pubs.acs.org/action/showLinks?pmid=17317294&crossref=10.1016%2Fj.bbalip.2007.01.007&coi=1%3ACAS%3A528%3ADC%252BD2sXovFensrs%253Dhttp://pubs.acs.org/action/showLinks?pmid=17317294&crossref=10.1016%2Fj.bbalip.2007.01.007&coi=1%3ACAS%3A528%3ADC%252BD2sXovFensrs%253Dhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fjm800733h&pmid=19053776&coi=1%3ACAS%3A528%3ADC%252BD1cXhsVWisb7Ihttp://pubs.acs.org/action/showLinks?pmid=18412436&crossref=10.1063%2F1.2829861&coi=1%3ACAS%3A528%3ADC%252BD1cXkvFyiu74%253Dhttp://pubs.acs.org/action/showLinks?pmid=16222654&crossref=10.1002%2Fjcc.20289&coi=1%3ACAS%3A528%3ADC%252BD2MXht1SlsbbJhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fct700119m&pmid=26636222&coi=1%3ACAS%3A528%3ADC%252BD2sXhtFaqsLzEhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fct700119m&pmid=26636222&coi=1%3ACAS%3A528%3ADC%252BD2sXhtFaqsLzEhttp://pubs.acs.org/action/showLinks?system=10.1021%2Fct700119m&pmid=26636222&coi=1%3ACAS%3A528%3ADC%252BD2sXhtFaqsLzEhttp://pubs.acs.org/action/showLinks?pmid=15888456&crossref=10.1074%2Fjbc.M502463200&coi=1%3ACAS%3A528%3ADC%252BD2MXlvFymu7o%253Dhttp://pubs.acs.org/action/showLinks?pmid=22244763&crossref=10.1016%2Fj.str.2011.10.018&coi=1%3ACAS%3A528%3ADC%252BC38XovF2qtw%253D%253Dhttp://pubs.acs.org/action/showLinks?pmid=23139175&crossref=10.1002%2Fpro.2190&coi=1%3ACAS%3A528%3ADC%252BC38XhvVeksrvM